In previous blogs and in several of our reports, we’ve covered the “three C’s of in-flight connectivity” (which should really be four when you consider the costs involved). Latency is another important, but often overlooked, part of the connectivity experience and is defined as the total time it takes a data packet to travel from one node to another. It is sometimes argued that latency has little bearing on most passenger-facing connectivity applications, and this may well be true in commercial aviation (although high latency can cause page load times to be slow when take rates are high). However, the way connectivity is used, and the expectations that accompany this use, are completely different in business aviation. Business travellers are much more inclined to use video conferencing software, have VoIP conversations and connect to a VPN. For each of these applications, latency is of paramount importance. Online in-flight gaming is another emerging application that can require a very low latency system. The rollout of 5G networks, which exhibit latency of between 20 and 30 milliseconds, will increase pressure on vendors to shorten the cycle time between the on-ground experience and expectations in the air.
According to NetForecast, an independent provider of broadband performance solutions, the average roundtrip packet time from a PED to an online service using a landline connection is 25 milliseconds. In-flight, however, across all currently deployed technologies, it is in the region of 790 milliseconds. Furthermore, the company estimates that packet loss, which is the number of packets that don’t make it to their destination and need to be re-sent, is around 0.05 per cent using a landline connection, but as high as 13 per cent on in-flight connections. Latency and packet loss at this level can, therefore, cause problems with web pages loading, especially if you have multiple users requesting data at the same time, creating a bottleneck that is independent of bandwidth.
While there are technological strategies to mitigate against the impact of latency on services, the only real way to minimise it is to reduce the distance between the origin of a data packet and its destination. For this reason, satellites in orbit at a higher altitude have a higher degree of latency than those in a lower orbit. The same is true of ATG communications. Because cell towers on the ground are closer to the aircraft flying above, latency is inherently lower than with any kind of satellite system. Another important consideration is the design of the connectivity system itself. Those that allocate the majority of their bandwidth in the forward link can expect to see a higher level of roundtrip latency than a symmetrical design where bandwidth is equally distributed between the forward and return link.
When it comes to satellite networks, it is also important to consider the impact of the ground network on latency. Tests of new LEO satellites have shown incredibly low latencies, but one should note that these are not necessarily representative of real-world conditions. OneWeb, for example, achieved average single trip latency of 32 milliseconds during testing in July 2019 and Telesat achieved 18 milliseconds round-trip latency in a February 2020 test. In both instances, there was no “true” ground network to speak of where a packet of data would travel from an aircraft to a satellite, to a ground station and an Internet breakout point (and back). Rather, these tests measured the physical round-trip time from terminal to ground (via satellite) but not out to the Internet via the ground network.
As most LEO networks are still in their infancy, their exists little data to show what average measured round-trip latency might look like on a business aircraft. We do know that whilst Iridium expects round-trip latency for its Certus solution to be in the region of 30 – 50 milliseconds in future, the network was actually pinging at about 500 milliseconds as of February 2019. Similarly, our understanding of OneWeb’s proposed architecture, had it been built out, is that round-trip latency could have been as low as 40 milliseconds or as high as 200 milliseconds, depending where in the world the aircraft happened to be and where traffic terminated on the ground. Along these lines, Telesat’s marketing material for its upcoming LEO constellation indicates that although round-trip latency for the space segment is expected to be less than 50 milliseconds, taking account of both the space and ground segments increases this to less than 100 milliseconds.
Furthermore, the Federal Communications Commission (FCC) recently provided information on why it doesn’t think SpaceX and can call itself low latency for purpose of getting funding under the bulk of the $16 billion rural broadband initiative. The proposal, released this week, is scheduled for a vote by the five-member commission at its 9th June meeting and suggests that – as intimated above – “the distance between Earth and satellites is not the only factor determining latency” and that “in the absence of a real world example of a non-geostationary orbit satellite network offering mass market fixed service to residential consumers that is able to meet our 100 millisecond round trip latency requirements, Commission staff could not conclude that such an applicant is reasonably capable of meeting our low latency requirements, and so we foreclose such applications”. SpaceX claims round-trip latency of its Starlink network will be less than 50 milliseconds.
MEO satellite networks are also in their infancy as far as their use in providing connectivity to business jets goes. SES, which does not yet use its O3b constellation for airborne connectivity, claims that general end-to-end round-trip latency is in the region of 140 milliseconds for data services. Likewise, we do not yet have an accurate read on what average round-trip latency will look like on a business jet connected to a next-gen ATG network such as those being developed by Gogo and SmartSky Networks. The latter, which will launch its network in 2020, one year ahead of Gogo’s new 5G ATG network, claims users will see round-trip latency below 100 milliseconds. Indeed, during various demo flights, the company has indicated that the latency when playing online multiplayer game, Fortnite, typically ranged between 70 and 90 milliseconds.
For these reasons, the table below shows only average measured round trip latencies for the two types of aircraft network commonly deployed today: legacy ATG and the GEO networks that have been the staple of satellite-based IFC for some time. For comparison, the table also shows what typical round trip latency looks like for familiar terrestrial networks such as home Internet and ground-based LTE.
Table 1: Comparison of Round-Trip Latency Associated with Different Networks